摘要
为提高筒子纱抓取和上纱过程的自动化和柔性化程度,应用单目视觉系统引导机器人完成上纱过程。采用工业相机获取单个筒子纱不同形态的数据信息,应用GAN(生成式对抗神经网络)对筒子纱数据集扩充,提高筒子纱数据集多样性。将所得的数据集加载到Faster R-CNN(更快速区域卷积神经网络)模型里进行训练,应用训练好的神经网络识别和定位筒子纱,引导机器人完成上纱任务。应用搭建的单目视觉系统实验平台对结果进行测试,结果表明,经过标定后的视觉系统可以完成多个筒子纱的抓取任务,以满足筒子纱上纱过程的自动化和柔性化要求。
In order to improve the automation and flexibility process in the process of gripping,the monocular vision system was used to guide the robot to complete the upper yarn process.Industrial cameras was used to obtain data information of different shapes of individual yarns,GAN was applied to generate neural network to expand the package data of the package yarn,and the diversity of the package yarn data set was improved.The obtained data set is loaded into the Faster R-CNN neural network model for training,and finally the trained neural recognition and positioning of the package yarn was guided to guide the robot to complete the task of yarn winding.Finally,the experimental platform of the monocular vision system was used to test the results.The experimental results show that the calibrated vision system can complete the task of grasping multiple packages,which can meet the automation and flexibility requirements of the yarn on the package.
作者
金守峰
林强强
马秋瑞
JIN Shoufeng;LIN Qiangqiang;MA Qiurui(College of Mechanical and Electrical Engineering,Xi'an Polytechnic University,Xi'an,Shaanxi 710600,China;College of Apparel&Art Design,Xi'an Polytechnic University,Xi'an,Shaanxi 710600,China)
出处
《毛纺科技》
CAS
北大核心
2020年第1期79-84,共6页
Wool Textile Journal
基金
陕西省自然科学基础研究计划项目(2017JM5141)
陕西省教育厅专项科研计划项目(17JK0334)
西安工程大学博士基金(BS1535)
西安工程大学研究生创新基金项目(chx2019083)
西安市科技局创新引导项目(201805030YD8CG14(5)
